Import and Export

Nicky Wakim

Importing data

  • We often need to work with a dataset in R
    • These data need to be brought into our environment so we can work with them!

 

  • We can import data from many file types, including .csv, .txt, .xlsx, .rds, .sas7bdat, and .dta

 

  • Once imported, R typically stores data as data frames (or tibbles) if using the {tidyverse} package

    • Usually use the term dataset to generally refer to either a data frame or tibble

Functions to import different file types

Here are various functions needed to import data from common file types:

File Type Extension(s) Function Package Example
CSV .csv read.csv(), read_csv() base R, readr read.csv("file.csv") , read_csv("file.csv")
Excel .xlsx, .xls read_excel() readxl read_excel("file.xlsx", sheet = 1)
RDS .rds read_rds() readr read_rds("file.rds")
SAS .sas7bdat read_sas() haven read_sas("file.sas7bdat")
Stata .dta read_dta() haven read_dta("file.dta")
SPSS .sav read_sav() haven read_sav("file.sav")

Let’s see an example with our HRS data

  • Notice I am using the here package to make a relative file path for my data

  • If my HRS dataset is a .rds file type

    • Note: I prefer .rds files when available because it can store factors and other R-specific information
    hrs_data_rds <- read_rds(here("./data/hrs_data.rds"))

 

  • If my HRS dataset is a .xlsx file type
hrs_data_xlsx <- read_excel(here("data", "hrs_data.xlsx"))

Glimpse at the data frame

tibble(hrs_data_rds) |> print(n = 7)
# A tibble: 2,728 × 32
  HHID   pn    id    BIRTHYR BIRTHMO BIRTHDATE proxy coupled sex   age_mo age_yr
  <fct>  <fct> <chr>   <dbl>   <dbl>     <dbl> <fct> <fct>   <fct>  <dbl>  <dbl>
1 552754 010   5527…    1967       9      2814 Resp… Not a … Fema…    665     55
2 555339 020   5553…    1966       7      2387 Resp… Couple… Fema…    676     56
3 559734 010   5597…    1967       3      2602 Resp… Couple… Male     663     55
4 558143 020   5581…    1973      10      5036 Resp… Couple… Male     600     50
5 554789 020   5547…    1950       7     -3153 Resp… Couple… Male     860     71
6 551018 010   5510…    1970       1      3667 Resp… Not a … Fema…    632     52
7 550516 010   5505…    1960       1        14 Resp… Not a … Male     749     62
# ℹ 2,721 more rows
# ℹ 21 more variables: ed <dbl>, degree <fct>, race_original <fct>,
#   smoke_ever <fct>, smoke_now <fct>, drink <fct>, height <dbl>, srh <fct>,
#   act_vig <fct>, bp <fct>, diab <fct>, cancer <fct>, lung <fct>, hrt <fct>,
#   strk <fct>, psych <fct>, sleep <fct>, arth <fct>, cond_count <dbl>,
#   cesd <dbl>, income <dbl>
tibble(hrs_data_xlsx) |> print(n = 7)
# A tibble: 2,728 × 32
  HHID   pn    id    BIRTHYR BIRTHMO BIRTHDATE proxy coupled sex   age_mo age_yr
  <chr>  <chr> <chr>   <dbl>   <dbl>     <dbl> <chr> <chr>   <chr>  <dbl>  <dbl>
1 552754 010   5527…    1967       9      2814 Resp… Not a … Fema…    665     55
2 555339 020   5553…    1966       7      2387 Resp… Couple… Fema…    676     56
3 559734 010   5597…    1967       3      2602 Resp… Couple… Male     663     55
4 558143 020   5581…    1973      10      5036 Resp… Couple… Male     600     50
5 554789 020   5547…    1950       7     -3153 Resp… Couple… Male     860     71
6 551018 010   5510…    1970       1      3667 Resp… Not a … Fema…    632     52
7 550516 010   5505…    1960       1        14 Resp… Not a … Male     749     62
# ℹ 2,721 more rows
# ℹ 21 more variables: ed <dbl>, degree <chr>, race_original <chr>,
#   smoke_ever <chr>, smoke_now <chr>, drink <chr>, height <dbl>, srh <chr>,
#   act_vig <chr>, bp <chr>, diab <chr>, cancer <chr>, lung <chr>, hrt <chr>,
#   strk <chr>, psych <chr>, sleep <chr>, arth <chr>, cond_count <dbl>,
#   cesd <dbl>, income <dbl>

rio package

  • The rio package provides a single, universal import/export function that works across many file types
  • Instead of remembering which package and function to use, import() detects the file type automatically from the extension
hrs_data_rds <- import(here("data", "hrs_data.rds"))
hrs_data_xlsx <- import(here("data", "hrs_data.xlsx"))
1
This will automatically use read_rds to import the dataset
2
This will automatically use read_excel to import the dataset
  • Great for quickly reading in data without worrying about the right package

 

Note: this is a new package! You will need to install it!

Optional: cleaning names (1/2)

  • Variable names in real datasets are often messy: spaces, capital letters, special characters
  • The clean_names() function from janitor standardizes all column names
names(hrs_data_rds) |> head(10)
1
Original names from our imported HRS dataset 
 [1] "HHID"      "pn"        "id"        "BIRTHYR"   "BIRTHMO"   "BIRTHDATE"
 [7] "proxy"     "coupled"   "sex"       "age_mo"   
hrs_data <- clean_names(hrs_data_rds)
names(hrs_data) |> head(10)
2
We can use clean_names() on the dataset directly
3
New, clean names for the HRS dataset 
 [1] "hhid"      "pn"        "id"        "birthyr"   "birthmo"   "birthdate"
 [7] "proxy"     "coupled"   "sex"       "age_mo"

 

  • Spaces are converted to underscores: “Age (years)” to “age_yrs”
  • Upper case letters are converted to lower case: “BIRTHDATE” to “birthdate”

Optional: cleaning names (2/2)

  • Consistent, clean names make coding easier and reduce errors

 

  • Can be piped directly after import:
hrs_data <- import(here("data", "hrs_data.rds")) |> clean_names()

Exporting data

  • After cleaning or modifying data, you may want to save it to a file
  • Use export() from rio to write data to any common format:
export(hrs_data, here("data", "hrs_data_clean.rds"))
export(hrs_data, here("data", "hrs_data_clean.csv"))
export(hrs_data, here("data", "hrs_data_clean.xlsx"))
1
Save a .rds file in the data folder
2
Save a .csv file in the data folder
3
Save a .xlsx file in the data folder
  • Prefer .rds for saving R objects between sessions: it preserves variable types exactly
  • Use .csv or .xlsx when sharing data with others outside of R

There are also the equivalent export functions

Here are various functions needed to export data from common file types:

File Type Extension(s) Function Package Example
CSV .csv write.csv(), write_csv() base R, readr write.csv(df, "file.csv"), write_csv(df, "file.csv")
Excel .xlsx write_xlsx() writexl write.xlsx(df, "file.xlsx")
RDS .rds write_rds() readr write_rds(df, "file.rds")
SAS .sas7bdat write_sas() haven write_sas(df, "file.sas7bdat")
Stata .dta write_dta() haven write_dta(df, "file.dta")
SPSS .sav write_sav() haven write_sav(df, "file.sav")

Wrap-up

  • Today we covered how to import and export data in R from a variety of file types

 

  • Key takeaways:
    • Use file-specific functions (read_csv(), read_excel(), etc.) or import() from rio for convenience
    • Always check variable types after importing with glimpse() or tibble()
    • Use clean_names() from janitor to standardize column names
    • Use export() or file-specific functions to save cleaned data

Resources